Skip to content

Conversation

@Valentine233
Copy link
Collaborator

@Valentine233 Valentine233 commented Jun 16, 2025

Use the manual transpose and pack instead of the onednn one, to get perf gain.
Depend on pytorch/pytorch#156065.

@pytorch-bot
Copy link

pytorch-bot bot commented Jun 16, 2025

🔗 Helpful Links

🧪 See artifacts and rendered test results at hud.pytorch.org/pr/pytorch/ao/2380

Note: Links to docs will display an error until the docs builds have been completed.

✅ No Failures

As of commit 313e6e3 with merge base 00417b8 (image):
💚 Looks good so far! There are no failures yet. 💚

This comment was automatically generated by Dr. CI and updates every 15 minutes.

@Valentine233 Valentine233 marked this pull request as draft June 16, 2025 07:47
@facebook-github-bot facebook-github-bot added the CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. label Jun 16, 2025
@Valentine233 Valentine233 added the topic: not user facing Use this tag if you don't want this PR to show up in release notes label Jun 16, 2025
pytorchmergebot pushed a commit to pytorch/pytorch that referenced this pull request Jul 7, 2025
@Valentine233 Valentine233 marked this pull request as ready for review July 7, 2025 03:06
@Valentine233 Valentine233 requested a review from jansel July 8, 2025 05:29
@Valentine233
Copy link
Collaborator Author

@drisspg @jerryzh168 @jansel Please help take a look to the PR.

@Valentine233 Valentine233 merged commit a45b1f7 into pytorch:main Jul 10, 2025
19 checks passed
liangel-02 pushed a commit that referenced this pull request Aug 25, 2025
* [cpu int8 sdpa] use manual tranpose and pack
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

CLA Signed This label is managed by the Facebook bot. Authors need to sign the CLA before a PR can be reviewed. topic: not user facing Use this tag if you don't want this PR to show up in release notes

Projects

None yet

Development

Successfully merging this pull request may close these issues.

4 participants